Development of the ReaxFF Reactive Force Field for Cu/Si Systems with Application to Copper Cluster Formation during Cu Diffusion Inside Silicon

Abstract

Transition-metal impurities such as nickel, copper, and iron in solid-state materials like silicon have a significant impact on the electrical performance of integrated circuits and solar cells. To study the impact of copper impurities inside bulk silicon on the electrical properties of the material, one needs to understand the configurational space of copper atoms incorporated inside the silicon lattice. In this work, we developed a ReaxFF reactive force field and used it to perform molecular dynamics simulations on models with up to 762 atoms to study the various configurations of individual and crystalline clusters of copper atoms inside bulk silicon by examining copper’s diffusional behavior in silicon. The ReaxFF Cu/Si parameter set was developed by training against density functional theory (DFT) data, including the energy barrier for an individual Cu atom traveling inside a silicon lattice. We found that the diffusion of copper atoms is dependent on temperature. Moreover, we show that individual copper atoms start to form clusters inside bulk silicon at temperatures above 500 K. Our simulation results provide a comprehensive understanding of the effects of temperature and copper concentration on the formation of copper clusters inside a silicon lattice. Finally, the stress–strain relationship of Cu/Si compounds under uniaxial tensile loading has been obtained. Our results indicate a decrease in the elastic modulus with increasing Cu-impurity concentration. We observe spontaneous microcracking of the Si during the stress–strain tests as a consequence of the formation of a small Cu cluster adjacent to the Si surface.

CunSi12 (n = 30, 38, and 60): A series of silicide cages with high content of TM atoms

“…For computational efficiency, DFT methods are very attractive and several studies have shown that they can be applied satisfactorily to many problems in main group and transition metal chemistry [32–35]. In addition, some results of DFT calculations are proved to match the experimental results quite well in Cu (Si)‐containing structures [36–40]. Moreover, the Cu 2 Si 2D monolayer featuring planar hexacoordinate copper, which was predicted to be stable on the basis of DFT computations [41], has been successfully prepared on Cu(111) [42] and Si(111) [43], respectively.…”

Development of the ReaxFF Reactive Force Field for Li/Mn/O Battery Technology with Application to Design a Self-Healing Cathode Electrolyte Interphase

“…34,35 In this method, Mn sites in the LiMn 2 O 4 spinel are replaced with other transition metals. 36,37 The doping of the spinel with Ni in the spinel structure reduces the vulnerability of Mn 3 + cations to acid attack, resulting in improved electrochemical performance and reduced dissolution of Mn. Nevertheless, extended cycling of nickel-doped spinel causes capacity degradation due to the formation of surface films.…”

Optimization of ReaxFF Reactive Force Field Parameters for Cu/Si/O Systems via Neural Network Inversion with Application to Copper Oxide Interaction with Silicon

“…Similarly, trace copper oxide contamination during the process of compound semiconductor fabrication can lead to catastrophic failures in the device performance. A significant concern for the semiconductor industry is the effectiveness of wafer cleaning. , Wet chemical cleaning is the preferred method to remove organic materials, particles, metallic impurities, and native silicon oxide from the wafer surface. − Therefore, it is critical to minimize the effect of impurities in chemicals used in the device-production lines. Urata et al studied the interaction between copper and amorphous silica (a-SiO 2 ) as a standard metal/oxide interaction by using reactive molecular dynamic (MD) simulations to understand the intrinsic role of an oxidized metal layer on the adhesion between a metal and an oxide glass .…”

“…Moreover, the limited number of search iterations makes it drastically faster than the brute force algorithm. 60 Previously, we developed a Cu/Si ReaxFF reactive force field and used it to perform molecular dynamics simulations 15 and to study the various configurations of individual and crystalline clusters of copper atoms inside bulk silicon by examining copper's diffusional behavior in silicon. In this work, we extend our study by developing a Cu/Si/O ReaxFF reactive force field describing Cu, O, and Si atom interactions by using neural network inversion as the default searching algorithm inside the INDEEDopt framework.…”